Comparison of mode split model based on multinomial logit and nested logit

The selection of traffic mode is related to the service level of the traffic mode, the characters of travelers and the characters of the trip. In order to reduce the forecasting error in traditional model, ML and NL mode split model are established in this paper. All the traffic modes are classified into one arrangement in ML model. While in NL model, the traffic modes that have similar characters are classified into the same arrangement. Discrete factors that influence the selection of traffic mode are age, occupation, income, travel objective, having a car or not, payment mode, continuous factors are travel time and the cost. According to the results, NL mode split model is more conformable to describe mode split behavior than ML model.